Skip to main content

markdown-footnote-length.hs

Linter that warns about overly long footnotes in Markdown files

Pathbuild/markdown-footnote-length.hs
LanguageHaskell
Lines29

Read this when Use this page when tracing the Haskell build pipeline, generators, metadata code, or backend utility behavior around markdown-footnote-length.

Overview​

This is a standalone lint script that scans Markdown files for footnotes exceeding a character threshold. It's part of gwern.net's quality assurance tooling, helping maintain readable footnote lengths.

The script parses Markdown using Pandoc, extracts all footnote blocks, converts them back to plain text to measure their rendered length, and prints warnings for any that exceed 2,400 characters. This threshold was chosen empirically—footnotes longer than this typically overflow the visible screen area on gwern.net's layout.

The tool is designed to be run via find across the entire wiki directory during content review or as part of build-time linting.

Public API​

CLI Usage​

runghc markdown-footnote-length.hs <file.md>

# Typical batch usage:
find ~/wiki/ -name "*.md" -exec runghc markdown-footnote-length {} \;

Arguments: Single Markdown file path

Output: For each footnote exceeding 2,400 characters, prints <filepath>: <footnote-content> to stdout

Exit: Returns non-zero on parse failures or missing args; otherwise exits 0 (warnings only)

Internal Architecture​

Processing Pipeline​

File path
    │
    ▼
TIO.readFile ─────► Raw Markdown Text
    │
    ▼
readMarkdown ─────► Pandoc AST (or parse error)
    │
    ▼
bottomUpM ────────► Traverse AST, checking each Inline
    │
    ▼
footNoteCheck ────► For Note nodes: convert to plain text, check length
    │
    ▼
TIO.putStrLn ─────► Print warning if > 2400 chars

Key Functions​

FunctionPurpose
mainEntry point, reads file from args
processLintParses Markdown, initiates AST traversal
footNoteCheckChecks individual footnote length

Data Flow​

  1. Parse: readMarkdown with full pandocExtensions converts Markdown to Pandoc AST
  2. Traverse: bottomUpM walks the AST bottom-up, applying footNoteCheck to each Inline node
  3. Check: When a Note inline is found, its block content is rendered back to plain text via writePlain
  4. Warn: If plain text length > 2,400 characters, print the file path and footnote content
Key Patterns

Bottom-Up Monadic Traversal​

The script uses Pandoc's bottomUpM for AST traversal, which processes children before parents. For footnote checking, traversal order doesn't matter—each Note is independent—but bottomUpM provides a convenient way to visit every Inline node with monadic effects (IO for printing warnings).

bottomUpM (footNoteCheck f) x'

Plain Text Rendering for Length Measurement​

Rather than measuring the raw Markdown source (which includes syntax characters), the script renders footnote content to plain text first. This gives a more accurate measure of the displayed length:

let md = runPure $ writePlain def (Pandoc nullMeta cntnts)
Configuration

Length Threshold​

ConstantValueLocationPurpose
Character limit2,400Hardcoded in footNoteCheckMaximum footnote length before warning

The threshold is not configurable via command line—it's embedded in the source at line 27:

when (T.length md' > 2400) $ ...

To change it, modify the source directly.

Integration Points​

Build System​

  • Not invoked by sync.sh or the main build pipeline
  • Invoked per-page by build/markdown-lint.sh as one of its lint checks
  • Also suitable for manual/periodic batch review via find -exec

Dependencies​

  • Pandoc library: For Markdown parsing and plain text rendering
  • Text.Pandoc.Definition: Inline(Note) pattern for footnote matching

Input/Output​

  • Input: Any .md file with Pandoc-compatible footnotes
  • Output: Warnings to stdout (not stderr), no file modifications
  • Side effects: None beyond console output
See Also